For the millions of adults living with knee osteoarthritis or recovering from joint replacement, meaningful monitoring of gait mechanics has long been locked inside expensive motion-capture laboratories. A methodology that moves this assessment into real-world settings could reshape how clinicians detect incomplete recovery and guide rehabilitation decisions after knee arthroplasty.
This study tested whether inertial measurement units (IMUs) — small wearable sensors measuring acceleration and angular velocity — could reliably estimate two clinically critical loading metrics during walking: the knee flexion moment and the knee adduction moment. These waveforms characterize the "stiff-knee" gait pattern common in osteoarthritis, where range of motion is reduced and stance-phase loading is more uniform. Using principal component analysis to distill the complex moment waveforms into interpretable pattern variables, researchers found that shank-mounted IMUs explained 55–64% of variance (R² = 0.55–0.64, p < 0.01) in both moment patterns across patients with end-stage knee OA. Adding foot-mounted sensors raised model performance to R² = 0.67–0.69, with agreement analyses showing minimal systematic bias, though moderate limits of agreement were noted.
These findings occupy a meaningful middle ground in wearable biomechanics research. The R² values achieved are modest but clinically promising — comparable to or exceeding prior IMU-based attempts to estimate joint kinetics without force plates. The knee adduction moment in particular is a well-established surrogate for medial compartment loading and a predictor of OA progression, making its ambulatory estimation especially valuable. That said, several important caveats apply: the cohort represents end-stage OA awaiting arthroplasty, limiting generalizability to earlier disease stages or post-surgical populations where gait variability differs substantially. Model derivation in a single-center laboratory environment also means external validation in community or home settings remains an essential next step. This work is best characterized as rigorous proof-of-concept rather than a clinical-ready tool, but it meaningfully advances the case that wearable gait kinetics can escape the laboratory.